Talk:XML ITS R1 schema tooling issues
After posting of 2006-10-05, Rene Spronk e-mailed me. His comments, and my in-line responses are below:
(Rene)I'm not a pubs facilitator, some shooting-from-the-hip comments: once a domain goes normative the database is never updated anymore by a committee. Therefore the problem will/may continue to exist until the domain goes out fro ballot again.
- (Woody) True, this is why I hoped it had been referred to I&M for correction of both ballot and Edition content.
- (Rene) As far as I'm aware that hasn't happened, but as I stated I'm not a pubs facilitator, so it may have gone via that channel. The problem is that the schema are not normative so any discrepancies would have to be noticed by implementers. I've got no idea where their comments would end up; they tend to create a workaround and not report the issue..
(Rene)*why are the VSDs not "auto saved" to the database each cycle? Most of them have one MT nowadays and it forces a synchronization to take place.
- (Woody)This would be a good idea if it were not for the fact that some committee's edit their constraints in the Repository-HMD, and still others create multiple HMDs or multiple MTs from a single RMIM. Therefore, the publishing wonks cannot do this routinely without messing up content.
- (Rene) True, currently it can't be done automatically because of HMD level constraints. Would it be possible to detect the fact that an R-MIM has only one MT in its HMD? In that case at least those Visios could be automatically saved (again) to the repository as part of the pubs process. This also encourages committees to use a 1-1 relationship between an R-MIM and a MT.
(Rene) Once we have MIFs to express both the visual diagram as well as the serialized model this problem will go away.. as things are now we can never really be sure that the Visios match the HMD and/or match the model saved in the database.
- (Woody) Correct. This is why careful preview and editing is required. BTW, if a committee requests a one-time action to clean up a circumstance like this one, I would be happy to consider it. Does I&M wish to make this request?
- (Rene) The thing is, these R-MIMs have HMDs with 2 message types (one of two possible payload stubs). INM probably won't use HMD-level constraints in future (we rarely use them anyway) - this makes the fix a manual process. We are however talking about a total of 3 RMIMs for the MFMI domain. So if you could help to fix this and avoid the issue from occuring in future Normative Editions I'd certainly appreciate it.